ipip: percpu stats accounting
authorEric Dumazet <eric.dumazet@gmail.com>
Mon, 27 Sep 2010 00:35:50 +0000 (00:35 +0000)
committerDavid S. Miller <davem@davemloft.net>
Tue, 28 Sep 2010 04:30:43 +0000 (21:30 -0700)
Maintain per_cpu tx_bytes, tx_packets, rx_bytes, rx_packets.

Other seldom used fields are kept in netdev->stats structure, possibly
unsafe.

This is a preliminary work to support lockless transmit path, and
correct RX stats, that are already unsafe.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/ipv4/ipip.c

index babd2527810694f28507da734ea3f3db7e665d7d..12b6fde6f65a3b3559dc8dee5fe17e29237fbe6a 100644 (file)
@@ -131,8 +131,9 @@ struct ipip_net {
        struct net_device *fb_tunnel_dev;
 };
 
-static void ipip_tunnel_init(struct net_device *dev);
+static int ipip_tunnel_init(struct net_device *dev);
 static void ipip_tunnel_setup(struct net_device *dev);
+static void ipip_dev_free(struct net_device *dev);
 
 /*
  * Locking : hash tables are protected by RCU and RTNL
@@ -141,6 +142,34 @@ static void ipip_tunnel_setup(struct net_device *dev);
 #define for_each_ip_tunnel_rcu(start) \
        for (t = rcu_dereference(start); t; t = rcu_dereference(t->next))
 
+/* often modified stats are per cpu, other are shared (netdev->stats) */
+struct pcpu_tstats {
+       unsigned long   rx_packets;
+       unsigned long   rx_bytes;
+       unsigned long   tx_packets;
+       unsigned long   tx_bytes;
+};
+
+static struct net_device_stats *ipip_get_stats(struct net_device *dev)
+{
+       struct pcpu_tstats sum = { 0 };
+       int i;
+
+       for_each_possible_cpu(i) {
+               const struct pcpu_tstats *tstats = per_cpu_ptr(dev->tstats, i);
+
+               sum.rx_packets += tstats->rx_packets;
+               sum.rx_bytes   += tstats->rx_bytes;
+               sum.tx_packets += tstats->tx_packets;
+               sum.tx_bytes   += tstats->tx_bytes;
+       }
+       dev->stats.rx_packets = sum.rx_packets;
+       dev->stats.rx_bytes   = sum.rx_bytes;
+       dev->stats.tx_packets = sum.tx_packets;
+       dev->stats.tx_bytes   = sum.tx_bytes;
+       return &dev->stats;
+}
+
 static struct ip_tunnel * ipip_tunnel_lookup(struct net *net,
                __be32 remote, __be32 local)
 {
@@ -239,7 +268,7 @@ static struct ip_tunnel * ipip_tunnel_locate(struct net *net,
        if (parms->name[0])
                strlcpy(name, parms->name, IFNAMSIZ);
        else
-               sprintf(name, "tunl%%d");
+               strcpy(name, "tunl%d");
 
        dev = alloc_netdev(sizeof(*t), name, ipip_tunnel_setup);
        if (dev == NULL)
@@ -255,7 +284,8 @@ static struct ip_tunnel * ipip_tunnel_locate(struct net *net,
        nt = netdev_priv(dev);
        nt->parms = *parms;
 
-       ipip_tunnel_init(dev);
+       if (ipip_tunnel_init(dev) < 0)
+               goto failed_free;
 
        if (register_netdevice(dev) < 0)
                goto failed_free;
@@ -265,7 +295,7 @@ static struct ip_tunnel * ipip_tunnel_locate(struct net *net,
        return nt;
 
 failed_free:
-       free_netdev(dev);
+       ipip_dev_free(dev);
        return NULL;
 }
 
@@ -359,8 +389,10 @@ static int ipip_rcv(struct sk_buff *skb)
        const struct iphdr *iph = ip_hdr(skb);
 
        rcu_read_lock();
-       if ((tunnel = ipip_tunnel_lookup(dev_net(skb->dev),
-                                       iph->saddr, iph->daddr)) != NULL) {
+       tunnel = ipip_tunnel_lookup(dev_net(skb->dev), iph->saddr, iph->daddr);
+       if (tunnel != NULL) {
+               struct pcpu_tstats *tstats;
+
                if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
                        rcu_read_unlock();
                        kfree_skb(skb);
@@ -374,7 +406,11 @@ static int ipip_rcv(struct sk_buff *skb)
                skb->protocol = htons(ETH_P_IP);
                skb->pkt_type = PACKET_HOST;
 
-               skb_tunnel_rx(skb, tunnel->dev);
+               tstats = this_cpu_ptr(tunnel->dev->tstats);
+               tstats->rx_packets++;
+               tstats->rx_bytes += skb->len;
+
+               __skb_tunnel_rx(skb, tunnel->dev);
 
                ipip_ecn_decapsulate(iph, skb);
 
@@ -397,13 +433,12 @@ static int ipip_rcv(struct sk_buff *skb)
 static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
 {
        struct ip_tunnel *tunnel = netdev_priv(dev);
-       struct net_device_stats *stats = &dev->stats;
-       struct netdev_queue *txq = netdev_get_tx_queue(dev, 0);
+       struct pcpu_tstats *tstats;
        struct iphdr  *tiph = &tunnel->parms.iph;
        u8     tos = tunnel->parms.iph.tos;
        __be16 df = tiph->frag_off;
        struct rtable *rt;                      /* Route to the other host */
-       struct net_device *tdev;                        /* Device to other host */
+       struct net_device *tdev;                /* Device to other host */
        struct iphdr  *old_iph = ip_hdr(skb);
        struct iphdr  *iph;                     /* Our new IP header */
        unsigned int max_headroom;              /* The extra header space needed */
@@ -413,13 +448,13 @@ static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
        if (skb->protocol != htons(ETH_P_IP))
                goto tx_error;
 
-       if (tos&1)
+       if (tos & 1)
                tos = old_iph->tos;
 
        if (!dst) {
                /* NBMA tunnel */
                if ((rt = skb_rtable(skb)) == NULL) {
-                       stats->tx_fifo_errors++;
+                       dev->stats.tx_fifo_errors++;
                        goto tx_error;
                }
                if ((dst = rt->rt_gateway) == 0)
@@ -427,14 +462,20 @@ static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
        }
 
        {
-               struct flowi fl = { .oif = tunnel->parms.link,
-                                   .nl_u = { .ip4_u =
-                                             { .daddr = dst,
-                                               .saddr = tiph->saddr,
-                                               .tos = RT_TOS(tos) } },
-                                   .proto = IPPROTO_IPIP };
+               struct flowi fl = {
+                       .oif = tunnel->parms.link,
+                       .nl_u = {
+                               .ip4_u = {
+                                       .daddr = dst,
+                                       .saddr = tiph->saddr,
+                                       .tos = RT_TOS(tos)
+                               }
+                       },
+                       .proto = IPPROTO_IPIP
+               };
+
                if (ip_route_output_key(dev_net(dev), &rt, &fl)) {
-                       stats->tx_carrier_errors++;
+                       dev->stats.tx_carrier_errors++;
                        goto tx_error_icmp;
                }
        }
@@ -442,7 +483,7 @@ static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
 
        if (tdev == dev) {
                ip_rt_put(rt);
-               stats->collisions++;
+               dev->stats.collisions++;
                goto tx_error;
        }
 
@@ -452,7 +493,7 @@ static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
                mtu = dst_mtu(&rt->dst) - sizeof(struct iphdr);
 
                if (mtu < 68) {
-                       stats->collisions++;
+                       dev->stats.collisions++;
                        ip_rt_put(rt);
                        goto tx_error;
                }
@@ -488,7 +529,7 @@ static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
                struct sk_buff *new_skb = skb_realloc_headroom(skb, max_headroom);
                if (!new_skb) {
                        ip_rt_put(rt);
-                       txq->tx_dropped++;
+                       dev->stats.tx_dropped++;
                        dev_kfree_skb(skb);
                        return NETDEV_TX_OK;
                }
@@ -525,14 +566,14 @@ static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
                iph->ttl        =       old_iph->ttl;
 
        nf_reset(skb);
-
-       IPTUNNEL_XMIT();
+       tstats = this_cpu_ptr(dev->tstats);
+       __IPTUNNEL_XMIT(tstats, &dev->stats);
        return NETDEV_TX_OK;
 
 tx_error_icmp:
        dst_link_failure(skb);
 tx_error:
-       stats->tx_errors++;
+       dev->stats.tx_errors++;
        dev_kfree_skb(skb);
        return NETDEV_TX_OK;
 }
@@ -547,13 +588,19 @@ static void ipip_tunnel_bind_dev(struct net_device *dev)
        iph = &tunnel->parms.iph;
 
        if (iph->daddr) {
-               struct flowi fl = { .oif = tunnel->parms.link,
-                                   .nl_u = { .ip4_u =
-                                             { .daddr = iph->daddr,
-                                               .saddr = iph->saddr,
-                                               .tos = RT_TOS(iph->tos) } },
-                                   .proto = IPPROTO_IPIP };
+               struct flowi fl = {
+                       .oif = tunnel->parms.link,
+                       .nl_u = {
+                               .ip4_u = {
+                                       .daddr = iph->daddr,
+                                       .saddr = iph->saddr,
+                                       .tos = RT_TOS(iph->tos)
+                               }
+                       },
+                       .proto = IPPROTO_IPIP
+               };
                struct rtable *rt;
+
                if (!ip_route_output_key(dev_net(dev), &rt, &fl)) {
                        tdev = rt->dst.dev;
                        ip_rt_put(rt);
@@ -699,13 +746,19 @@ static const struct net_device_ops ipip_netdev_ops = {
        .ndo_start_xmit = ipip_tunnel_xmit,
        .ndo_do_ioctl   = ipip_tunnel_ioctl,
        .ndo_change_mtu = ipip_tunnel_change_mtu,
-
+       .ndo_get_stats  = ipip_get_stats,
 };
 
+static void ipip_dev_free(struct net_device *dev)
+{
+       free_percpu(dev->tstats);
+       free_netdev(dev);
+}
+
 static void ipip_tunnel_setup(struct net_device *dev)
 {
        dev->netdev_ops         = &ipip_netdev_ops;
-       dev->destructor         = free_netdev;
+       dev->destructor         = ipip_dev_free;
 
        dev->type               = ARPHRD_TUNNEL;
        dev->hard_header_len    = LL_MAX_HEADER + sizeof(struct iphdr);
@@ -717,7 +770,7 @@ static void ipip_tunnel_setup(struct net_device *dev)
        dev->priv_flags         &= ~IFF_XMIT_DST_RELEASE;
 }
 
-static void ipip_tunnel_init(struct net_device *dev)
+static int ipip_tunnel_init(struct net_device *dev)
 {
        struct ip_tunnel *tunnel = netdev_priv(dev);
 
@@ -728,6 +781,12 @@ static void ipip_tunnel_init(struct net_device *dev)
        memcpy(dev->broadcast, &tunnel->parms.iph.daddr, 4);
 
        ipip_tunnel_bind_dev(dev);
+
+       dev->tstats = alloc_percpu(struct pcpu_tstats);
+       if (!dev->tstats)
+               return -ENOMEM;
+
+       return 0;
 }
 
 static void __net_init ipip_fb_tunnel_init(struct net_device *dev)